In general, if I have a choice between being right or wrong, I would rather be right. Certainly I tend to like life better when I am right (something my wife sarcastically notes from time to time). Unfortunately, every once in awhile, I say stuff that turns out to be wrong. At least, I say stuff that could be interpreted as being incorrect.
A good example is my coverage of the Milwaukee Bucks this past summer. Here are the two observations I made with respect to this team.
1. The Bucks are boring. Across the past two decades the Bucks, more than any other team, consistently hovered around the average mark. In other words, Milwaukee has not been a truly outstanding team. And they didn’t tend to fail miserably. This team has just been average, and we tend to think of average as boring.
2. The Bucks in 2007-08 will be bad. In reviewing the Eastern Conference I noted that the Celtics would be really good (got that right). The Bulls would be nearly as good (okay, that looks to be wrong also, but I will talk about that later in the week). The Pistons and Cavaliers will be good, but not as good as the Celtics (basically right). The next ten teams in the conference are going to be hard to rank (Orlando, Miami, and New York makes this look wrong, but many teams fit that description). And every team in the conference will be better than Milwaukee, because the Bucks will be bad. In fact, I thought the Bucks would probably only win about 25 games.
The Bucks Defy Prediction
This past weekend the Bucks defeated the Dallas Mavericks, one of the better teams in the Association. And now, after eleven games, the Bucks are 7-4. With a winning percentage of 64%, this team is on pace to win 52 games. This is not boring. And it certainly ain’t bad. How is this possible?
First of all, when we look at the team’s offensive and defensive efficiency, we don’t see a team on pace to win 50 games. The Bucks are scoring 103.7 points per 100 possessions while allowing 104.2 points. With an efficiency differential of -0.5, this team should expect to win about 39 games. Of course, that’s still better than 25 victories. In fact it might even be good enough to make the playoffs.
Moving from efficiency differential to Wins Produced, we can see which players are responsible for this improvement. Before I get to the players who are responsible, let me eliminate a few candidates people might suspect: Andrew Bogut, Charlie Villanueva, and Yi Jianlian.
Bogut posted a 0.145 WP48 [Wins Produced per 48 minutes] last year. This year his mark is 0.132. So he has not improved. He is still a bit better than average (average is 0.100), but he’s not getting any better.
Villanueva posted a WP48 of 0.051 last year. People expected that he could play better, but after 224 minutes this year his WP48 stands at -0.027. So he’s has gone from below average to very bad.
And then there is Yi, the team’s lottery pick in 2007. Yi has so far posted a -0.001 WP48, which is obviously below average. Table One reports his performance with respect to the individual statistics.
So far Yi is well below average with respect to shooting efficiency and assists. The good news is that if he starts hitting his shots and making an occasional pass, he could become at least an average NBA player. He certainly has demonstrated the ability to rebound at the rate of an average power forward. And he can block shots. Still, he needs to hit his shots if he is going to be effective. Since he has not so far, he can’t be the reason this team improved.
To see who is responsible, let’s look at two projections of Wins Produced for the Bucks. The first is based on what each player (except Yi) did on a per-minute basis last year. The second is based on what they have done this year.
When we project from last year’s performance we see why I was so pessimistic about this team. Given what these players did in 2006-07, this team would be projected to win about 27 games this season. The problem is that there was simply no major Wins Producers on the Bucks in 2006-07. When your best player only has a WP48 of 0.145, it’s hard to believe your team is going to be a major force in the league. And I didn’t think Yi was going to come in and be a major force his rookie season.
But one player, after eleven games, has been a major force. Michael Redd in 2006-07 posted a WP48 of 0.130. This year, with 13% of the season gone, Redd has a 0.269 WP48. Where did Redd improve?
Table Three reports Redd’s performance in 2006-07 and after eleven games this year. Relative to last year his shooting efficiency and shot attempts are down. And so his points are also lower. But his rebounds and assists have increased dramatically, and hence, despite less scoring, his Wins Production is much higher.
Although Redd has made the biggest leap, he’s not the entire story. At the bottom of Table Two is Desmond Mason. Here is what Mason has done each of the past three seasons.
2004-05: -0.004 WP48, -0.2 Wins Produced, 2,893 minutes played
2005-06: -0.047 WP48, -2.0 Wins Produced, 2,104 minutes played
2006-07: -0.044 WP48, -2.3 Wins Produced, 2,575 minutes played
When you look at this record you see a player who keeps getting minutes and keeps failing to produce. Remember that last year the Bucks had Ruben Patterson at small forward [0.166 WP48, 8.7 Wins Produced]. To move from a player who was this productive to one that has been consistently negative across three seasons certainly suggests the team is taking a step backward.
Somehow, though, after eleven games Mason has not been horrible. Yes, he is still below average. He just isn’t that far below average.
Where has he improved? Table Four reports what Mason did last year and after eleven games this season.
In Mason’s first seven seasons he was below average with respect to shooting efficiency every single year. This year, though, he is well above average. Okay, he’s only taken 90 shots. Had he made only 40, instead of 46, his WP48 would have fallen into the negative range. Yes he is better with respect to turnovers, but the key for Mason is that he has hit a few more shots than you would expect. As a result, he is on pace to produce 4.8 wins more than his performance last year would suggest.
Summarizing the Changes
Okay, let’s summarize. Given what Milwaukee’s players did last year, we would expect a team that would win fewer than 30 games. But the record for the Bucks after eleven games suggests this team is on pace to finish with 50 wins. Looking at what this team did, the difference between the current pace and what we would expect can be explained by three factors:
1. The team is getting more wins than their efficiency differential would suggest. The team’s efficiency differential suggests a team that is going to win about half its games.
2. If the Bucks win half their games, though, they are still beating the projection. This has been made possible primarily because Redd has focused on something other than just scoring.
3. And last (and probably least) Mason has hit a few more shots than we would expect.
Can this continue? I am more optimistic about Redd than I am about Mason. Mason has logged 16,811 minutes in the NBA and has never shown he can consistently hit his shots. I do not think 289 minutes in a player’s 8th season should be seen as the beginning of something new.
As for Redd, maybe he has changed a bit. Well, change is the wrong word. Maybe Redd has returned to what we saw in 2001-02 and 2002-03. In these seasons Redd grabbed more than seven rebounds per 48 minutes and posted a WP48 of 0.196 and 0.237 respectively. After the 2002-03 campaign, though, Redd began taking more and more shots and doing less and less rebounding. Perhaps someone told him it would be better to be a bit more rounded.
If Redd has returned to what we saw five years ago, then the Bucks might just contend for a playoff spot in the East.
And if the Bucks do make the playoffs, then I guess I was wrong. Of course should I have expected Redd to become what he was five years ago? I mean, come on. How the hell was I supposed to see that coming?
Oh well. It’s not like this is the first time I was wrong. Just look at the Bulls. What’s happened to that team? Hopefully I can get to that story later this week.
Our research on the NBA was summarized HERE.
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.